According to analysis, Li Ziyang, a fund manager in the equity investment department of Morgan Stanley Fund, stated that the AI industry chain has been experiencing robust momentum this year, extending from the initial computing power sector throughout the entire PCB and semiconductor supply chains.
This is driven by the high growth in AI demand, which is boosting prices and profitability across the supply chain, moving beyond just computing chips to upstream and supporting segments. In the second quarter, the entire semiconductor industry, from advanced to mature process nodes, is showing signs of supply shortages.
Against the backdrop of highly visible AI demand, leaders in memory, advanced processes, and advanced packaging are now showing a clear willingness to actively expand production capacity.
Looking ahead to the second half of the year, Li Ziyang noted that while there are market discussions and debates about an "AI bubble," excessive concern is unnecessary. Given the higher confidence in the commercialization of AI, the positive momentum in the AI industry chain is expected to continue.
Concurrently, expectations for overseas interest rate hikes persist. If subsequent overseas economic and employment data remain strong, and inflation falls slower than expected, liquidity may continue to tighten, potentially suppressing sector valuations and increasing market volatility.
Specifically regarding the computing power industry chain, with the explosive growth in token demand driven by coding and AI agents, revenue for large language models and cloud providers is entering a new phase of accelerated growth.
The industry has now established a positive feedback loop: continuous iteration of large models leads to steady growth in corporate revenue, which in turn fuels increased AI capital expenditure. This business model is gradually proving viable.
Domestic internet giants, led by ByteDance, have clearly announced significant increases in AI investment. Leading chip manufacturers have substantially raised their shipment forecasts for 2026.
Domestic large model developers continue to catch up effectively through engineering capabilities and cost-efficient models, with leading companies beginning to gradually release strong earnings.
Li Ziyang mentioned that the strength of both overseas and domestic computing power supply chains continues to exceed expectations. Structurally tight segments, including optical communication and memory, remain strong with widening supply gaps.
Furthermore, inflationary pressures within the chain are spreading to more areas such as CPUs, electronic components, and upstream PCBs and semiconductor materials.
Regarding the semiconductor industry chain, benefiting from increased demand for advanced process chips and memory driven by artificial intelligence, the global semiconductor cycle continues its stronger-than-expected upswing.
Memory is beginning to see large-scale, long-term price-lock agreements with cloud providers, which is favorable for valuation increases.
According to the latest forecast from the World Semiconductor Trade Statistics (WSTS) organization, the global semiconductor market size is expected to reach $1.51 trillion by 2026.
Long-term equipment procurement plans by global industry leaders are locking in sustained demand. The semiconductor equipment sector is poised for a strong, two-year upcycle from 2026 to 2027.
The overall assessment is that the semiconductor sector remains in the middle phase of its current cycle. Investors may focus on targets where there is an expectation gap and which have not yet been fully priced in by the market.
Comments